Sorting LiteLLM GitHub issues by difficulty to implement. Higher = harder.

{
Issue #16060 is harder because it involves diagnosing state, concurrency, filtering, and fallback behavior across the router’s usage-based deployment-selection machinery. Issue #31149 appears narrower, centered on propagating a resolved endpoint through one Anthropic token-counting request path, with related reports likely providing a focused reproduction and fix boundary.
}
https://github.com/berriai/litellm/issues/16060 3:2 https://github.com/berriai/litellm/issues/31149